The Home That Guesses vs. The Home You Already Told

IFA wrapped up in Berlin on 8 September. It's the world's largest consumer electronics and home appliance trade fair, and this year's theme was "The Future is Now," with more than 1,900 brands from 49 countries on the floor.
Euronews sent a reporter to walk the halls and answer one question: what does the home of the future actually look like? Their answer was interesting, and it's worth unpacking, because it lands right on top of an argument I make to clients every week.
The industry is moving from "smart" to "intelligent"
The framing Euronews used was that the industry is shifting from the "smart home," where a device responds to a command, to the "intelligent home," where a device learns, adapts, and anticipates what you need before you ask.
The examples were clever. Haier showed a washing machine with a built-in camera that can see what's in the drum and flag it if a red sock has snuck in with the white load, before it ruins your shirt. Neil Tunstall, Haier Europe's CEO, told Euronews: "The product now will do that for you. And that hopefully means then that we can actually enjoy eating much more at home." He was talking about ovens that read what's inside them and adjust the cook, but the point applies across the stand: less thinking for the person, more thinking done by the appliance.
A US company called Cozyla took it further, demonstrating a system that coordinates a household's schedules, meal planning, and daily routines across a family, from morning to bedtime. TCL, meanwhile, showed the opposite instinct: technology designed to help you disconnect, like a phone mode that strips out notifications so you can actually read.
None of this is nonsense. A washing machine that catches a colour-run mistake is a useful, low-risk piece of intelligence, and it stays contained. One appliance, one outcome, easy to check and easy to undo if it gets it wrong.
Guessing works fine until the guess is wrong
Here's where I think the framing gets shaky. Euronews' own closing line was that the vision is an ecosystem where devices "increasingly anticipate what needs to be done," so people spend less time managing individual devices and more time on things they actually want to do.
That's a lovely idea for a sock. It's a different proposition entirely when you're talking about the lighting, the blinds, the HVAC, and the security system for an entire house.
An appliance guessing wrong costs you a pink shirt. A whole-home system guessing wrong closes your blinds mid-video call, kills the lights while you're still in the room, or decides you've "left" the house when you've just stepped onto the patio. The bigger the surface area a system tries to anticipate, the more ways it has to get it wrong, and the less forgiving those mistakes become.
This is the distinction I'd draw for any client looking at this year's IFA coverage and wondering what it means for their own build: device-level intelligence, a machine making a narrow, reversible call about its own narrow job, is worth having. Whole-home orchestration built on the same anticipate-and-guess model is solving the wrong problem.
Precision, not prediction
A KNX system doesn't try to read you. It doesn't need a camera, a behavioural model, or a cloud service deciding what it thinks you probably want tonight. You tell it, once, exactly what "movie night" means for your lights, your blinds, your AV rack, and your climate. You tell it what "leaving the house" means for the alarm, the geyser, and the irrigation. That gets programmed into the bus through ETS, and from then on it runs precisely what you specified, the same way, every single time, whether it's the first time or the thousandth.
The house doesn't need time to learn your habits, and it never quietly decides it knows your intent better than you told it. If you want the scene to change, you change the programming yourself. It doesn't relearn on its own.
That precision matters more here than it does in markets with rock-solid grid and connectivity. An anticipatory, AI-driven system depends on a live connection to whatever model is doing the anticipating. When the power blips or the router drops, that intelligence has nothing left to work with, and it either falls back to a dumb default or simply stops. A KNX scene doesn't care whether the internet is up. It runs off the bus, locally, exactly as programmed, load shedding or not.
What this means at the brief stage
I'm not telling clients to avoid smart appliances that do their one job better. A washing machine that catches a colour mistake, an oven that reads what's inside it, these are welcome, low-stakes upgrades and I'd never talk anyone out of them.
What I am saying is that the system responsible for running the actual house, the one an architect or designer needs to spec at the brief stage, shouldn't be built on the same "let the AI guess" logic as a laundry appliance. Whole-home control needs to be deterministic: specified once, on an open standard, running on hardware that doesn't need permission from a server somewhere to turn a light on.
IFA's vision of the intelligent home makes sense for what it's trying to do. It's just aimed at a different, narrower problem than the one my clients are solving when they brief a smart home from scratch. Let the oven guess what temperature your dinner needs. Don't let anything guess whether your front door is locked.
Just a conversation about what the project needs.
wayne@knxlogic.co.za | 082 564 3982 | www.knxlogic.co.za




Comments